(NOTE: Cross-posted on DB2-L) We have a java application which is showing rapid garbage collection/allocation failure loops (every 10-15 seconds) when using the IBM DB2 UDB JDBC Drivers, we have seen this in IBMJava2 SR2 and SR5. (Linux is SuSE SLES7; DB2 V7.2)
Basically when we issue a query (i.e. select * from db2inst1.tablename) we immediately begin to see a rapid succession of Allocation Failures and then Garbage Collections, we are not certain that this is a bad thing or not. It would make sense if this were some type of buffering mechanism for the query, allocate what you can to end-of-memory, fill it with results, display it, garbage collect it, lather, rinse, and repeat. The issue is that the amount of memory washed through during this process far exceeds the amount of data in the table by several orders of magnitude (example: 150MB table goes through approx 6GB of memory). I don't know if this is a bug or just working as designed/coded, but in either case it seems entirely inefficient. Thanks, Mike
